This is an automated email from the ASF dual-hosted git repository.
tlopex pushed a commit to branch main
in repository https://gitbox.apache.org/repos/asf/tvm.git
The following commit(s) were added to refs/heads/main by this push:
new be2ea89e44 [Docs] Update outdated references from recent refactors
(#18906)
be2ea89e44 is described below
commit be2ea89e443f5eb857ce1217d0287a076c0b0fbe
Author: Shushi Hong <[email protected]>
AuthorDate: Mon Mar 16 12:07:23 2026 -0400
[Docs] Update outdated references from recent refactors (#18906)
- Update Python version requirement from 3.7/3.8 to 3.10 in docs
- Migrate lint commands from `docker/lint.sh` / `docker/bash.sh ci_lint`
to `pre-commit` (#18807)
- Replace `make` / `make runtime` with CMake commands (root Makefile
removed in #18828)
- Remove `dmlc::ThreadLocalStore` reference in pass_infra (dmlc phased
out in #18779)
- Update FFI header links in runtime.rst (`packed_func.h` →
`tvm/ffi/function.h`)
- Remove stale `tvm.testing.ErrorTest` example in error_handling.rst
- Remove dead `apps/extension` link in runtime.rst
---
docs/arch/pass_infra.rst | 4 --
docs/arch/runtime.rst | 18 ++++-----
docs/contribute/code_guide.rst | 14 +++----
docs/contribute/error_handling.rst | 44 ++++------------------
docs/contribute/pull_request.rst | 35 +++++------------
docs/how_to/tutorials/cross_compilation_and_rpc.py | 9 +++--
docs/install/from_source.rst | 4 +-
7 files changed, 40 insertions(+), 88 deletions(-)
diff --git a/docs/arch/pass_infra.rst b/docs/arch/pass_infra.rst
index a1019c88ad..8826251ba8 100644
--- a/docs/arch/pass_infra.rst
+++ b/docs/arch/pass_infra.rst
@@ -162,10 +162,6 @@ Python APIs to create a compilation pipeline using pass
context.
}
};
- /*! \brief The thread-local store to hold the pass context. */
- typedef dmlc::ThreadLocalStore<PassContextThreadLocalEntry>
- PassContextThreadLocalStore;
-
Pass Constructs
^^^^^^^^^^^^^^^
diff --git a/docs/arch/runtime.rst b/docs/arch/runtime.rst
index 99c83de837..72b47705ec 100644
--- a/docs/arch/runtime.rst
+++ b/docs/arch/runtime.rst
@@ -48,7 +48,7 @@ function call whose caller and callee may be in different
languages.
The following code block provides an example in C++
-.. _PackedFunc:
https://github.com/apache/tvm/blob/main/include/tvm/runtime/packed_func.h
+.. _PackedFunc:
https://github.com/apache/tvm/blob/main/3rdparty/tvm-ffi/include/tvm/ffi/function.h
.. code:: c
@@ -138,7 +138,7 @@ which allows us to embed the PackedFunc into any languages.
Besides python, so f
`java`_ and `javascript`_.
This philosophy of embedded API is very like Lua, except that we don't have a
new language but use C++.
-.. _minimum C API:
https://github.com/apache/tvm/blob/main/include/tvm/runtime/base.h
+.. _minimum C API:
https://github.com/apache/tvm/blob/main/3rdparty/tvm-ffi/include/tvm/ffi/c_api.h
.. _java: https://github.com/apache/tvm/tree/main/jvm
.. _javascript: https://github.com/apache/tvm/tree/main/web
@@ -257,20 +257,18 @@ Each argument in PackedFunc contains a union value
`TVMValue`_
and a type code. This design allows the dynamically typed language to convert
to the corresponding type directly, and statically typed language to
do runtime type checking during conversion.
-.. _TVMValue:
https://github.com/apache/tvm/blob/main/include/tvm/runtime/base.h#L135
+.. _TVMValue:
https://github.com/apache/tvm/blob/main/3rdparty/tvm-ffi/include/tvm/ffi/c_api.h
The relevant files are
-- `packed_func.h`_ for C++ API
-- `c_runtime_api.cc`_ for C API and how to provide callback.
+- `function.h`_ for C++ PackedFunc API
+- `c_api.h`_ for C API.
-.. _packed_func.h:
https://github.com/apache/tvm/blob/main/include/tvm/runtime/packed_func.h
-.. _c_runtime_api.cc:
https://github.com/apache/tvm/blob/main/src/runtime/c_runtime_api.cc#L262
+.. _function.h:
https://github.com/apache/tvm/blob/main/3rdparty/tvm-ffi/include/tvm/ffi/function.h
+.. _c_api.h:
https://github.com/apache/tvm/blob/main/3rdparty/tvm-ffi/include/tvm/ffi/c_api.h
To support extension types, we used a registry system to register type related
information, like support of any
-in C++, see `Extension types`_ for more details.
-
-.. _Extension types: https://github.com/apache/tvm/tree/main/apps/extension
+in C++. See the ``tvm-ffi`` subproject under ``3rdparty/tvm-ffi/`` for more
details on the FFI type system.
Runtime-Specific Information
diff --git a/docs/contribute/code_guide.rst b/docs/contribute/code_guide.rst
index ad09610ac9..28d61c5053 100644
--- a/docs/contribute/code_guide.rst
+++ b/docs/contribute/code_guide.rst
@@ -41,15 +41,15 @@ C++ Code Styles
We use ``clang-format`` to enforce the code style. Because different version
of clang-format might change by its version, it is recommended to use the same
version of the clang-format as the main one.
-You can also use the following command via docker.
+You can use pre-commit hooks to run formatting checks:
.. code:: bash
- # Run a specific file through clang-format
- docker/bash.sh ci_lint clang-format-10 [path-to-file]
+ # Run clang-format on all files
+ pre-commit run clang-format --all-files
- # Run all linters, including clang-format
- python tests/scripts/ci.py lint
+ # Run all linters (Python + C++ + custom checks)
+ pre-commit run --all-files
clang-format is also not perfect, when necessary, you can use disble
clang-format on certain code regions.
@@ -86,8 +86,8 @@ Because clang-format may not recognize macros, it is
recommended to use macro li
Python Code Styles
------------------
- The functions and classes are documented in `numpydoc
<https://numpydoc.readthedocs.io/en/latest/>`_ format.
-- Check your code style using ``python tests/scripts/ci.py lint``
-- Stick to language features in ``python 3.7``
+- Check your code style using ``pre-commit run --all-files``
+- Stick to language features in ``python 3.10``
- For functions with early returns, prefer ``if``/``elif``/``else``
chains for functions with parallel and short bodies to the
diff --git a/docs/contribute/error_handling.rst
b/docs/contribute/error_handling.rst
index a6049419ad..c580977ddb 100644
--- a/docs/contribute/error_handling.rst
+++ b/docs/contribute/error_handling.rst
@@ -55,43 +55,13 @@ The following code gives an example on how to do so.
}
}
-The above function is registered as PackedFunc into the python frontend,
-under the name ``tvm._api_internal._ErrorTest``.
-Here is what will happen if we call the registered function:
-
-.. code::
-
- >>> import tvm
- >>> tvm.testing.ErrorTest(0, 1)
- Traceback (most recent call last):
- File "<stdin>", line 1, in <module>
- File "/path/to/tvm/python/tvm/_ffi/_ctypes/function.py", line 190, in
__call__
- raise get_last_ffi_error()
- ValueError: Traceback (most recent call last):
- [bt] (3) /path/to/tvm/build/libtvm.so(TVMFuncCall+0x48) [0x7fab500b8ca8]
- [bt] (2) /path/to/tvm/build/libtvm.so(+0x1c4126) [0x7fab4f7f5126]
- [bt] (1) /path/to/tvm/build/libtvm.so(+0x1ba2f8) [0x7fab4f7eb2f8]
- [bt] (0) /path/to/tvm/build/libtvm.so(+0x177d12) [0x7fab4f7a8d12]
- File "/path/to/tvm/src/api/api_test.cc", line 80
- ValueError: Check failed: x == y (0 vs. 1) : expect x and y to be equal.
- >>>
- >>> tvm.testing.ErrorTest(1, 1)
- Traceback (most recent call last):
- File "<stdin>", line 1, in <module>
- File "/path/to/tvm/python/tvm/_ffi/_ctypes/function.py", line 190, in
__call__
- raise get_last_ffi_error()
- tvm.error.InternalError: Traceback (most recent call last):
- [bt] (3) /path/to/tvm/build/libtvm.so(TVMFuncCall+0x48) [0x7fab500b8ca8]
- [bt] (2) /path/to/tvm/build/libtvm.so(+0x1c4126) [0x7fab4f7f5126]
- [bt] (1) /path/to/tvm/build/libtvm.so(+0x1ba35c) [0x7fab4f7eb35c]
- [bt] (0) /path/to/tvm/build/libtvm.so(+0x177d12) [0x7fab4f7a8d12]
- File "/path/to/tvm/src/api/api_test.cc", line 83
- InternalError: cannot reach here
- TVM hint: You hit an internal error. Please open a thread on
https://discuss.tvm.ai/ to report it.
-
-As you can see in the above example, TVM's ffi system combines
-both the python and c++'s stacktrace into a single message, and generate the
-corresponding error class automatically.
+When a C++ function registered via the FFI raises an error with a typed prefix,
+the TVM FFI system will automatically map it to the corresponding Python
exception
+class. For example, a ``ValueError:`` prefix in the error message will raise a
Python
+``ValueError``, and an ``InternalError:`` prefix will raise
``tvm.error.InternalError``.
+
+TVM's FFI system combines both the Python and C++ stacktraces into a single
message,
+and generates the corresponding error class automatically.
How to choose an Error Type
diff --git a/docs/contribute/pull_request.rst b/docs/contribute/pull_request.rst
index d7e0bc2ab4..c20be2df38 100644
--- a/docs/contribute/pull_request.rst
+++ b/docs/contribute/pull_request.rst
@@ -39,25 +39,13 @@ Guidelines
.. code:: bash
- # While the lint commands used should be identical to those run in CI,
this command reproduces
- # the CI lint procedure exactly (typically helpful for debugging lint
script errors or
- # to avoid installing tools manually)
- python tests/scripts/ci.py lint
+ # Run all lint checks via pre-commit hooks
+ pre-commit run --all-files
- # Run all lint steps.
- docker/lint.sh
-
- # To run steps individually, specify their step names on the
command-line. An incorrectly
- # spelled step name causes the tool to print all available steps.
- docker/lint.sh <step_name> ...
-
- If the clang-format lint check fails, run git-clang-format as follows to
automatically reformat
- your code:
-
- .. code:: bash
-
- # Run clang-format check for all the files that changed since
upstream/main
- docker/bash.sh ci_lint ./tests/lint/git-clang-format.sh --rev
upstream/main
+ # Run specific linters individually
+ pre-commit run ruff-check --all-files # Python lint
+ pre-commit run ruff-format --all-files # Python format
+ pre-commit run clang-format --all-files # C++ format
- Add test-cases to cover the new features or bugfix the patch introduces.
- Document the code you wrote, see more at :ref:`doc_guide`
@@ -260,8 +248,8 @@ If you want to run all tests:
.. code:: bash
- # build tvm
- make
+ # build tvm (see install-from-source for CMake build instructions)
+ cd build && cmake .. && cmake --build . --parallel $(nproc) && cd ..
./tests/scripts/task_python_unittest.sh
@@ -269,14 +257,11 @@ If you want to run a single test:
.. code:: bash
- # build tvm
- make
-
# let python know where to find tvm related libraries
export PYTHONPATH=python
rm -rf python/tvm/*.pyc python/tvm/*/*.pyc python/tvm/*/*/*.pyc
- TVM_FFI=ctypes python -m pytest -v
tests/python/unittest/test_pass_storage_rewrite.py
+ python -m pytest -v
tests/python/tir-transform/test_tir_transform_storage_rewrite.py
# Additionally if you want to run a single test, for example
test_all_elemwise inside a file.
- TVM_FFI=ctypes python -m pytest -v -k "test_all_elemwise"
tests/python/frontend/tflite/test_forward.py
+ python -m pytest -v -k "test_all_elemwise"
tests/python/frontend/tflite/test_forward.py
diff --git a/docs/how_to/tutorials/cross_compilation_and_rpc.py
b/docs/how_to/tutorials/cross_compilation_and_rpc.py
index 0e6346619a..e4383278e8 100644
--- a/docs/how_to/tutorials/cross_compilation_and_rpc.py
+++ b/docs/how_to/tutorials/cross_compilation_and_rpc.py
@@ -52,7 +52,9 @@ and the Firefly-RK3399 for an OpenCL example.
#
# git clone --recursive https://github.com/apache/tvm tvm
# cd tvm
-# make runtime -j2
+# mkdir build && cd build
+# cp ../cmake/config.cmake .
+# cmake .. && cmake --build . --parallel $(nproc)
#
# After building the runtime successfully, we need to set environment variables
# in :code:`~/.bashrc` file. We can edit :code:`~/.bashrc`
@@ -215,9 +217,10 @@ print(f"{cost:g} secs/op")
#
# .. code-block:: bash
#
-# cp cmake/config.cmake .
+# mkdir -p build && cd build
+# cp ../cmake/config.cmake .
# sed -i "s/USE_OPENCL OFF/USE_OPENCL ON/" config.cmake
-# make runtime -j4
+# cmake .. && cmake --build . --parallel $(nproc)
#
# The following function shows how we run an OpenCL kernel remotely
diff --git a/docs/install/from_source.rst b/docs/install/from_source.rst
index fd423fa7a1..d84185912d 100644
--- a/docs/install/from_source.rst
+++ b/docs/install/from_source.rst
@@ -40,7 +40,7 @@ Apache TVM requires the following dependencies:
- Clang 5.0
- Apple Clang 9.3
- Visual Studio 2019 (v16.7)
-- Python (>= 3.8)
+- Python (>= 3.10)
- (Optional) Conda (Strongly Recommended)
System Dependencies (Non-Conda)
@@ -339,4 +339,4 @@ tests in TVM. The easiest way to install GTest is from
source.
make
sudo make install
-After installing GTest, the C++ tests can be built and started with
``./tests/scripts/task_cpp_unittest.sh`` or just built with ``make cpptest``.
+After installing GTest, the C++ tests can be built and started with
``./tests/scripts/task_cpp_unittest.sh``, or built via CMake with
``-DUSE_GTEST=ON`` and then running ``./build/cpptest``.
